Maish Saidel-Keesing: Revolutionizing Deployment: System Initiative Meets Containers

In this talk, the speaker discusses the evolution and impact of deployment systems, specifically focusing on the System Initiative and its relationship with container technology and Amazon Web Services (AWS). He explains the significance of containers in modern application architecture as a standard unit of deployment and describes how Amazon ECS serves as a managed platform for running containers effortlessly. The speaker also highlights the concept of platform engineering, introducing it as a practice derived from DevOps principles to enhance developer experience and streamline workflows. Although the speaker initially intended to demonstrate the System Initiative tool, he notes challenges in its current state, emphasizing the importance of continuous adaptation and feedback from users to improve technology solutions over time.

[Music] ladies and Gentlemen please welcome our next speaker M sadal keing presenting the topic revolutionizing deployment system initiative meets containers um thank you very much for that wonderful AI introduction the name was almost there but my name is m CLE casing but fine no worries perfect um I'm actually very very happy to be here it's actually my first time in Lithuania um and yesterday when I was

walking around doing a little bit of touring I was um trying to get back to my roots and actually it's quite funny because yes my family was in Lithuania from the late 18th century till almost the second world war born and died here um I was born in a completely different country in South Africa if you know a little bit about the Jewish history that's pretty much

what happened a lot of the Jews left from Lithuania to go to South but still um there's a lot of history here and the thing about history is that it's nice and familiar and comfortable we know we're talking about and when you encounter something which is old and looks familiar to you or something which you um identify with it it gives you a good feeling and on

the other case we also love modern things um we like doing things in a new way learning new stuff and the reason we try to do things in a more modern way is to get things done as fast as possible with as little effort as possible spending as as little money as we can that's the way I look at it and specifically in technology so um how's

this all connected to the talk you'll see now approximately 10 months ago I encountered a blog post something about called a technology or new protocol initiative and we'll talk a little bit about more what the actual product is and I said wow this looks wonderful it's right at my alley it works with the products that I'm working with in my company in AWS and I'll propose a

talk and the talk got accepted 10 months ago approximately or eight months ago and since then I kind of left that on the back burner and didn't do very much about it until I started working in with stock um I like the product at the time it looked very nice looked very interesting but not everything turned out the way it was and that we'll see further down

the road exactly how how that actually turned out was an all smoooth sailing and we will'll go through the Journey kind of together so what originally was planned to be a flashy demo and the wonderful thing that everything works perfectly is not going to be in stock because that's not what happened but I'm going to talk a little bit more about tooling on containers and a little

bit about AWS and we'll see how that goes so as I said my name is midle casing I'm a developer advocate for one of the service teams in AWS I I assume you know what AWS is um I hope so if not we're a cloud company and provide a lot of um different services that our customers can use located all over the world my role as a

developer Advocate is to sometimes do what I'm doing today and that is to come up and speak on stage but also to speak to you attendees and customers to hear your feedback about our specific products and my specific product which that I'm embedded in one of the service teams in ECS elastic container services to get your feedback to hear your pain points and see how we can

improve our products directly with part of the service team instead of going through the whole chain of sales or account teams that you might have I have direct access to the engineers who are writing the software which you use daily so what is our agenda for today a tiny bit about what system initiative is a little bit about containers um elastic container service ECS and more about

platform engineering which we'll dive a little bit more deep into and see how um goes so system initiative system initiative is an open source product uh project founded by two people Adam Jacobs and I apologize if I am butchering his name on camera um where is my Mouse um Mah lupinacci I hope it's I hope I said it correctly um so let me go back there for

a second and it was we all know what devops was supposed to be devops was the vision of getting developers and operators or people running systems to work together properly in one streamlined team or in a way with using tooling that they could do that properly and system initiative was founded as a tool again as a tool to help that process along the way but we all

know I hope at least this is what I feel is that devops is not and probably this will be a heal that I will might die on one day it's not a person it's not a team and it's not a tool devops is a culture or a cultural change inside your organization that you need to go through in order to streamline your workloads and your and deployments

and they came up with this tool as I said is one of its tool to help you with that process okay so yeah we're back so containers is anybody in the room using containers is anybody in the room not containers can I ask you why not yet okay so the answer from the audience was that they're not there yet and that's why they're here today also to

learn about these things but I'm hopefully very sure that you're going to learn a lot more today and tomorrow not only about containers but modern or applications and a lot of other things as well but containers have pretty much become the de facto standard of deploying the unit of deployment today in modern application architecture it's the reason being is because as you can see on the ship

they're all pretty much the same in other words from the outside what's inside the container can be whatever you pretty much want and it can be if we're talking about the analogy of containers on the ship it can be toys or food or clothing but the outside box is a standard thing which you can use and stack them in this very specific way which makes it very

easy to ship offload onload and deal with and move things around pretty much the same thing with containers containers has a standard wrapper with a open standard which everybody use and you can use that actual container to do a lot of things and run it whenever you want most important thing it's pretty much works everywhere and when we talk talking about working everywhere it means it works

on your laptop windows or Linux or Mac or wherever it may be and you can it out of your laptop onto physical servers your on premium infrastructure into a cloud or many clouds because the same package unit is the same and it works pretty much everywhere if it is on your kubernetes cluster if it's on your ECS cluster the same uniform container will still work of course

there's wer wrappers around that you need to put inside and configuration in order to make that actually easily adapt to the very specific deployment and environment you're dealing with but the actual unit everywhere else which is wonderful we all know the wonderful meme of it worked on my machine so this pretty much solves issue again not 100% because there's a lot of rappers depending on where you're

deploying but the actual unit will actually work and you don't have to worry about that so lot of you might be um familiar with AWS and ECS and if you're not let me just explain Amazon ECS is elastic container service it's a native container service running on AWS itself um and when we talk about a native container service it's also managed container service in other words it's

manage control plane that you don't have to touch all you have to do is interact with it with your API with your client into the API and it runs you don't pay for that you just pay for the compute nodes that you actually run your workloads in your containers in the cloud and they can be deployed on ec2 instances which are virtual machines or on a serverless

compute platform called AWS fargate where we take in AWS more of the operational overhead of running your compute infrastructure in the cloud and if you also want to we have options for running closer out of the cloud it'ss ECS anywhere or on the edge there's different options you can use for the same apis the same infrastructure the same way of methodology of deployment and when we talk

about scale it's an Enterprise scale Solutions we take care of and this was numbers from reinvent last year which was November so I assume it has been updated since then but still I'll use the numbers I don't have anything else at the moment we launch a task as a container we launch more than two and a quarter billion tasks every single week across all of our regions

tens of thousands of API requests per second across every single AWS region and the interesting thing is that customers that come to AWS 65% of all new customers want to do something in containers if it be in ECS or in eks in kubernetes elastic kubernetes service but the customers are ready to move into those kind of um mindset and that kind of technology so let's talk a

little bit about platform engineering for a bit has anybody heard this term before engineering I would say about half of the audience here in the room and I hope some of you also which are watching remotely put your hand up if not that's okay um so I was looking for a definition of platform engineering and I came across two different things highlighted small points which I wanted

to hear in in green over here this is from the Microsoft documentation on their website in fact it's a kind of a practice built from devops principles improving developer experience it's also a product and also a set of tools and systems in order to achieve it the second one is from a gentleman by the name of Luca garante who is the CEO or font of a company

which for the life of me I can't remember what it was called at the moment so we'll leave it at that Luca um and it's pretty much the same thing it's a discipline of Designing tools and workflows and um some kind of culture in order for you to manage your application on a regular basis and the question I would like to pose today is this sounds very

familiar this is what we were talking about devops exactly the same thing so are we kind of Reinventing the wheel again like we usually do on a regular basis just calling It Something New by looking for providing a solution looking for a problem instead of finding a a solution to sorry providing a problem instead and by giving a solution instead of that we should be actually providing

a solution for an existing problem which already exists it's a question which I will'll leave open-ended for you to think about and um I don't really have an answer at the moment um something which I have to think about and see how this actually evolves but to me it looks like it's a bit of a perhaps a hype which is currently in the market which might evolve

and might become something bigger who knows nobody can really look into the future but we'll have to see um I assume a lot of you under know what the cncf landscape is well this is not it this is the cncf landscape is approximately 15 times more condensed with logos and icons on it this is only the specific platform engineering landscape so yes there are so many tools

and again it moves into different layers and unfortunately um you can see it's a set of tools essentially that's what platform engineering is but the thing is why do we need it and the answer I think is very very simple because we want something simple to do at least your developers and your engineers who are writing your software want to not deal with all the plumbing underneath

they want to write their code commit their code and have it so if we could achieve that that would amazing but it's not always that easy it's complicated very very complicated and again it is not only as I said it's not only as we talked about devop before it's not a tool it's not a person it's not a team it's a combination and culture change and all

those three together creating some kind of a platform or platform engineering team or a platform engineering tool which we a lot of people are calling today developer portal is something which a lot of companies want because they don't want to deal with the plumbing as we said but it's very difficult because because essentially it's stitching a lot of different kind of tools together we had our landscape

before which was all of these things this is what your platform engineering tools team your platform engineering team does they make these things somehow work together and it's not as if you can get something off the box out off the shelf from a specific company because every single company is different some of them use GitHub Enterprise on Prem some of them use GitHub in the cloud and

not exactly the same some use bit bucket or gitlab or some other um Source control mechanism and different kind of workflow workflow injures if they have cicd processes or tools every single company different because of your own history and if you're starting new then you have the wonderful um privilege of being able to create everything from scratch and have no history but most people do have that

history in the background we said we were two people here who put up the hand that they're not using containers because haven't got there because I assume you're working for a company that still has history in the background and you're not there yet that history usually makes a lot of money a huge amount of money so it's not so simple to move and not so simple to

move quickly so the goal of all of them is to create something what do you want to do you want to make something that you will love and when I say say you will love I mean specifically me who's using it and if you're creating it you have to be that user as well other words I'm not going to create a product or a platform or developer

portal for somebody which I will never use you have to continue speaking to your customers and your customers in this case are your engineers and developers in order for them to understand you have to understand their pain of what their problems that they're experiencing and understand how you can improve and make their lives easier what tools are they missing what kind of um information do they need

to have in order for them to complete their job in a quicker way in Amazon we have a process or a mechanism called um working backwards working backwards is a process which we speak with all of our customers to understand what they want or what they're experiencing in order to dive deep into the problem of what they're what they're actually having at the moment and how they

would like to try and solve it we can come up with a solution which is specific for them or give them insight into Solutions which other have solved but the whole idea is to get the customer to focus exactly on what the problem is because if you say Okay I want to deploy software faster that's wonderful but what is stopping you oh because I have five different

tools which are not working together okay so what is stopping you from rewriting those things um my teams don't talk to each other so then you have to start getting your team there no no software or solution will ever be able to solve problems with which are not connected with actual the actual pain your and in the end all that they want is something which is simple

a simple way to deploy their software or test their software and iterate in a very very fast manner in order for them to deliver the business value to the your customers that's all that company wants that's all any company wants is in order for you to devel deliver that value and the way they can do it is by doing things in a fast and safe and secure

standard way so that might be by deploying and allowing them to deploy certain templates and certain standard infrastructure for them to do but always remember that you need to leave yourself with certain flexibility in order to adapt to the future why am I going backwards customers often come to us and ask where do I start and they will be different for every single person for every customer

we talk to what I like to do when I speak to customers is always focus on easing your pain and every single one of you know your problems in your environment and where things are not working correctly and you know exactly what bothers you the most and maybe keeps you up in the middle of the night because something is working I tend to um suggest to customers

to focus on the really really hard Parts first for a number of reasons firstly because when it's solved and you working on it properly will solve the problem number one you'll feel very very proud and very happy that you managed to do it secondly your business will love you because you managed to solve a very big problem and third by solving those problems it will allow you

to take the what they call the easy apples off the tree the very simple things which have already been solved before it will make you it will allow you to invest further with all the other things which were maybe not as important but will be easy to solve or change based on the changes that you have made in order for you to continue your journey and gain

further credibility in your team and help yourself but you do have to be flexible and when I mean flexible it means nobody can predict the future well maybe there are people but I I specifically definitely do not predict the future um I can guarantee you 100% or 99.9 whatever 5 119 however many you want that six months ago not a single pipeline or developer platform or devops

team whatever you would like to call call it was interested in gen not today people were starting to think of how generative AI can help them with their cicd P pipeline for example can it review pull requests automatically do those tests on its own and if it does find a bug automatically based on some kind of generative AI model look for a solution or present a possible

solution to the developer who's getting the error in the logs to help them along their way nobody would have thought about this six months ago if they did they were very very clever but I don't know anybody that did we can't predict the future I don't know what's going to happen in a year or in two years down the road but what you should be thinking about

is trying to allow yourself as much flexibility as you can for the future to adapt and adopt new things things will change they change every six months every one year every two years every five years they change the way we worked in the IT industry approximately 5 years ago is completely different to what we're doing now there's no nothing at all completely nothing at all similar in

any way the tools that I use today are not the tools that I use 5 days 5 years ago not even 6 months ago to be honest so we had continuously adapt and we change when I talk about planning for future I got actually we were walking around yesterday as um speakers we went for sorry about that um we were working around yesterday the speakers after we

had dinner thank you very much by the way for the organizers which were organizing for us for for for that and we came across this church and I think it's called the St an's church if I'm correct is that right yeah and from what I understand the legend goes and this what what the tour guy told us yesterday that the there was a master architect or designer

or engineer who um started with a product and he said okay we'll build the first floor and then he was called away from some emergency or some kind of other project that he needed to and he let apprentices continue to work and they added on some microservices on top of monolith and they continued to work and added even more microservices on top of them on our left

so whatever you plan in the beginning of course from According to the legend as far as I understood when they finished the building the master architect took them to the top floor and threw them off but I hope nobody will get killed for adapting and changing over um over time with your uh with your products with your Solutions but things will change whatever you thought about starting

in the beginning not necessarily will be what you end up with at all continue to Benchmark on a regular your applications your platforms you don't want to become the bottleneck of everybody who's trying to deliver your software which means you need to speak to your customers on a regular basis and again your customers are the people software find out where they're in where they're in they they're

encountering bottlenecks of any kind and um see how you can improve on a regular basis adapt change because change is the only thing which is constant I will try and do a demo but I can't promise Because the Internet wasn't working and with a what Hotpot from my phone is also not working very well but still okay so system initi I did promise as part of the

the abstract demos so I'd like to go back to the original story of talking about the fact of familiar and running after the new thing so when I looked at firstly this is a screenshot of what system initiative is it's a UI tool which allows you to integrate and drag and drop different pieces on your on a canvas and connect them together and with an apply change

it will deploy and up all your infrastructure on a regular basis from um in your AWS account when I originally T tried this tool approximately 10 months ago after seeing this blog post which was the integration where they added a module for Amazon ECS said well this is going to be cool going to try it out I did at the time you got a managed environment provided

to you as a trial for 10 days and you could set up the environment and start working I did it looked kind of cool bit slow at the time but um I said okay this is good enough for me to present a talk about it um I'll play with it more in the future and goes fast forward about seven or eight months down the road I didn't

really touch it very much then it's not to say that the project isn't active it's actually very very active it's very uh they have regular um commits on GitHub they have regular sync and update meetings on YouTube which they stream live for the team in order for their progress reports so they're actually very active and are doing a lot of good work and a lot of changes

have been have been happening since the previous first time that I touched the the product um a good couple of months ago and when I try to deploy it now for the talk there is no managed environment so the managed environment that I had once upon a time no longer exists you have to do everything locally and that means deploying everything on your laptop and there are

very specific instructions of how to do this and it works it is not easy sorry it is very easy but it's not it's a lot of moving Parts in the background that extracted everything into one command to bring up a development environment but you don't to understand that under the underlying infrastructure takes a lot of um a little bit more of knowledge and deep um acquaintance with

these kind of Technologies but still I tried to do that on my Mac my is 3 years four years old I'm still waiting for a refresh that's fine um but it ran but my fans shot up because it used a huge amount of resources I couldn't do anything else besides run this demo and I said I can't really for example run the demo on screen and present

at the same time so that's not going to work so I deployed the same software as an ec2 instance inside the cloud in AWS in my account a beefy beefy instance we're talking about something like 72 CPUs 64 GB of RAM a lot a very big machine and because everything is built from scratch everything is compiled at runtime so this project is still in a very very

early stage it is changing on a regular basis for example they rewrote the whole back end engine approximately 3 weeks ago so it takes time it runs pretty much okay it's not easy so my original idea of presenting a wonderful demo of container with system initiative will not really happen because it's to be perfectly honest it's not a tool that you should be using in production and

I would not yes to play around with perhaps but the tool is not yet ready there's a lot of problems with inside the tool that need to be addressed before somebody can actually say okay I would use this as a tool to deploy my own um my own own solution just as a matter of Interest they only got to the Milestone of using their own tool to

deploy their own software in other words the whole environment of development and production three weeks ago so it takes time and iteration and even though this was really a cool idea in the beginning as a demo I'm not going to do that because um I don't think it would be useful for you to invest too much time of trying to move everything over to this product because

it still has to evolve a little bit more and of course if you are interested it's an open source project so please feel free to participate in the Discord where they have a very very very friendly community of people willing to help answer questions if you have any pull requests that you would like to submit to change software you can also do that as um what happened

there and my screen went yeah three resources um first thing of course is the link for the system initiative on the web um goes to the main site you can learn a little bit little more about the actual product of course it take you to the GitHub repo the second one is a landing page for ECS where um you can learn more about containers and things that

I do on a day-to-day basis third and last one is is the platform engineering link which is a platform engineering org where you can um learn more about platform engineering the tools which are involved you can get a much better version of the landscape um screenshot that I placed it on my um on my deck over here so you can see exactly what we're talking about my

name is m that's my email and Twitter handle um I unfortunate flying out later this afternoon so I won't be here for tomorrow but um if you have any questions I would love to talk to you on back containers devops H I would say pretty much anything except for politics and religion because I don't think this will go anywhere good um but if you want to talk

about anything technology wise I'd be more than happy to sit down for the time that I have here for the rest of afternoon um and if not you can always reach me either by my email email or ping me on Twitter I'd be more than happy to answer um anything or talk about anything technology wise and um before we close I would just like say there was

a couple of questions which came in from the Q&A uh there are no questions in slido but if you have any I can bring the microphone you can use this opportunity and ask is anybody with questions I promise you I don't bite or you can Shout if was the question are the other products which do similar things okay so the question was are there other kind of

products or Solutions or Open Source Products which do these kind of things as well yes they are they're in the top of that um Square on the Bott of the developer portal one which is very very popular today is something called backstage which integrates very very well with kubernetes with actually have and I specifically did a live stream on this 3 weeks ago about solution which we

have where allows backstage to integrate with not only cetes but also with ECS and with Lambda and a number of other products which are we have all kinds of modules which allow you to use that same developer portal to deploy those kind of things on many services within AWS it's called Opa not open policy agent it's Opa on AWS if you look for that somewhere on the

link on the uh internet you'll find out what we're talking about but a lot of companies which we are speaking with customers are using something called backstage which seems to be the most popular at the moment it's not the only one again will depend on the specific um use case that you have your organization your culture any other questions it has been my honor and privilege to

speak here today thank you very very much for inviting me thank you thank you so much and see the ask me anything Corner yeah ask there's ask me anything corner somewhere outside I understand it's near the checkin near the check-in I'll be there until approximately 5:00 and have to start making my way back to go back home um thank you very much for joining today and I

hope it was useful session